Stochastic Fluid Models for conununication networks

A natural modeling framework for packet-based communication networks is provided through discrete event systems and, in particular, queueing models. However, the huge traffic volume that networks are supporting today makes such models highly impractical. An alternative modeling paradigm is based on Stochastic Fluid Models (SFM). The SFM paradigm allows the aggregation of multiple events, associated with the movement of individual packets over a time period of a constant flow rate, into a single event associated with a rate change. Using SFMs for the purpose of control and optimization rather than just performance analysis it has been recently shown that Perturbation Analysis (PA) techniques lead to simple unbiased sensitivity estimators that can be used not only in a simulation environment but also on line, driven by actual network data and requiring no knowledge of the traffic processes involved.